2
$\begingroup$

I asked a similar question earlier (How is the fiber alignment done in pluggable optical transceiver module (e.g., QSFP-DD)?), but few things are still not clear to me, so I'm posting another one here.

So far, my understanding is that a single-mode optical fiber (circular waveguide) has two degenerate modes. Due to this, even if a transmitter sends signals in TE mode (most of the rectangular silicon waveguide on the SOI platform is driven in TE mode), the fiber's output has unknown polarizations, assuming the fiber does not have a polarization-maintaining feature.

Let's assume that a given optical transceiver deals with a single polarization-sensitive scheme. The optical fiber's output polarization is unknown. Still, the silicon photonics-based receiver must take light from the fiber and couple it to the silicon waveguide (again, only TE mode is supported). Now my question is, what is the best way to couple the light from fiber to the silicon photonics-based system without using polarization-maintaining fiber (PMF)?

Especially the part that I am not convinced about is the coupler part. Suppose a fiber's polarization is unknown, but the silicon waveguide can only take TE mode. Doesn't the coupling loss vary significantly depending on how the fiber is attached to the system and the external stress applied to the fiber? Suppose 50% of light in fiber is in TE mode and the other 50% of light in fiber is in TM mode (even if Tx sends a signal in TE, due to the circular shape of a fiber, I believe the mode can be easily coupled back and forth between TE and TM, even for single-mode fiber). How can a coupler receive light from a fiber without too much loss? And if the fiber suddenly experiences more physical stress and now TE is 10%, and TM is 90%, how does the optical receiver make sure coupling loss is maintained without PMF?

$\endgroup$

1 Answer 1

2
$\begingroup$

You are right. Without polarization maintaining fiber, the polarization presented to the coupler could initially be anything. What’s done (or at least, what I’ve personally done in a research setting) is to use one of the polarization controllers with the paddles, and adjust it to optimize the coupling into the waveguide. Then you can assume that the polarization is the optimum one. You have to make sure the fiber is secured in place because, as you point out, the polarization can change if the fiber moves around. But if you, say, tape the fiber to the table and use PM fiber for most of the loose sections, it’s pretty stable.

So, unfortunately, it’s all very empirical. Just set it up, see what you get, and tweak things to optimize. But this arrangement has been enough to get integrated photonics as far as it is.

$\endgroup$
3
  • $\begingroup$ thank you for your comment. On top of your response, I also did extra research and found good enough reasoning/answers to this question. $\endgroup$
    – Emm386
    Commented Sep 13, 2022 at 22:06
  • $\begingroup$ @Emm386 did you find differing answers elsewhere? $\endgroup$
    – Gilbert
    Commented Sep 14, 2022 at 0:54
  • $\begingroup$ Your response is 100% right. one of other reason I asked this question is that some researchers place 2D grating coupler for polarization diversity purposes, but some don't. It turns out that those researchers not placing polarization diversity grating because they used external polarization controllers using paddles on the optical table. In addition, pluggable optical transceiver module often uses a bulky optical component that only includes fiber couplers and Ge PDs in a separate package. I am 100% convinced that this separate package includes a polarization splitter/rotator internally. $\endgroup$
    – Emm386
    Commented Sep 14, 2022 at 17:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.